In a message dated 11/03/2005, jpopelish@........ writes:

(snip  advice on wiring, circuit board and resistor thermoelectric
effects)  
> You can bolt a Cu chip cover strip onto the board. 
> This is  preferable to trying to reduce the dissipation by
> reducing the supply  voltage.

How so?  Preferable for what reasons?
Hi John,
 
    You are at most going to make a 50% reduction in  the power, which is 
better controlled by efficient heat sinking. What you  definitely do not want to 
do is to operate near the lower voltage limit with  degraded performance and 
less 'headroom'.

>  Seismometer amplifiers often have two distinct gain stages, with
> a  high pass filter set to maybe 20 to 30 sec in between. This will
>  greatly reduce thermal error signals and 1/f noise at the output.

I am  looking at using an integrator as negative feedback from the
output back to  the second or third stage to actively null the output
drift, instead of  putting a large capacitor in the forward signal
path.  This reduces  the net offset to the offset of the integrator,
instead of being the  amplifier offset of the input where a series
capacitor would normally be  used.  I think it is also easier to get a
low noise, low frequency  high pass corner frequency this way.  Have
you any criticism of this  technique?


Yes. You will still see the large 1/f deviations at  the output, which are of 
no interest and contain no information. You may find it  difficult to make it 
stable over several opamps, especially if the gain is  variable. If you want 
a sharper cutoff, why not use a two pole high pass  filter?
 
    Regards,
 
    Chris Chapman





In a message dated 11/03/2005, jpopelish@........ writes:
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>(snip=20
  advice on wiring, circuit board and resistor thermoelectric
effects)=20
  
> You can bolt a Cu chip cover strip onto the board. 
> This=20=
is=20
  preferable to trying to reduce the dissipation by
> reducing the sup=
ply=20
  voltage.

How so?  Preferable for what reasons?
Hi John,
 
    You are at most going to make a 50% reduction i=
n=20
the power, which is better controlled by efficient heat sinking. What you=20
definitely do not want to do is to operate near the lower voltage limit with=
=20
degraded performance and less 'headroom'.
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>>=20
  Seismometer amplifiers often have two distinct gain stages, with
> a=
=20
  high pass filter set to maybe 20 to 30 sec in between. This will
>=20
  greatly reduce thermal error signals and 1/f noise at the output.

I=
 am=20
  looking at using an integrator as negative feedback from the
output bac=
k to=20
  the second or third stage to actively null the output
drift, instead of=
=20
  putting a large capacitor in the forward signal
path.  This reduce=
s=20
  the net offset to the offset of the integrator,
instead of being the=20
  amplifier offset of the input where a series
capacitor would normally b=
e=20
  used.  I think it is also easier to get a
low noise, low frequency=
=20
  high pass corner frequency this way.  Have
you any criticism of th=
is=20
  technique?


    Yes. You will still see the large 1/f deviation=
s at=20
the output, which are of no interest and contain no information. You may fin=
d it=20
difficult to make it stable over several opamps, especially if the gain is=20
variable. If you want a sharper cutoff, why not use a two pole high pass=20
filter?
 
    Regards,
 
    Chris Chapman